🚀 AI Battery Management System (AI BMS) – Battery Performance Optimization

---

🌟 Project Overview

This project is part of the AI Battery Management System (AI BMS) initiative. The goal is to develop a real-time mode selection interface that allows users to optimize battery performance, efficiency, and longevity.

Users can choose between several operational modes, including:

• ⚡ Performance Mode
• 🌱 Eco Mode
• ⚖️ Balanced Mode
• 🛠️ Custom Mode

Modes dynamically adjust parameters such as cooling, fan speed, and temperature settings.

---

📊 User Interface Prototype

Below is a prototype of the AI Battery Management System (AI BMS) optimization interface:

Key Features of the Interface:

1. Mode Selection:

   • Users can choose between:
     • ⚡ Performance Mode: Maximize performance at the potential expense of battery life.
     • 🌱 Eco Mode: Optimize energy efficiency and extend battery life.
     • ⚖️ Balanced Mode: A middle ground between performance and efficiency.
     • 🛠️ Custom Mode: Configure user-defined parameters.

2. Custom Mode Settings:

   • Options to adjust:
     • Cooling Mode (Normal, Silent, Aggressive)
     • Fan Speed via slider
     • Flow Rate via slider
     • Temperature Threshold via slider
   • Enable/Disable advanced settings:
     • ✅ Overheat Protection
     • ✅ Fast Charging

3. Impact Visualizations:

   • Performance Impact: Bar chart showing mode-specific performance effects.
   • Efficiency Impact: Highlights energy efficiency in each mode.
   • Longevity Impact: Demonstrates how modes affect battery lifespan.

4. Interactive Controls:

   • Apply, Save, Cancel buttons for user actions.
   • Real-time visualization of the selected mode's impact.

---

✨ Features

• 🔄 Mode Switching: Users can select operational modes that dynamically adjust system parameters like fan speed and cooling temperature.
• 🛠️ Custom Mode: Users can personalize parameters such as fan speed and temperature thresholds, and see real-time effects on battery performance.
• 📈 Dynamic Updates: Real-time graphs display the impact of mode changes on performance, efficiency, and battery longevity.
• 🤖 Auto Adjustments: AI models automatically switch modes based on operating conditions.
• 🚗 Applications: Used in electric vehicles and energy storage systems to optimize battery performance in real time.

---

🛠️ Technologies Used

• Programming Language: 🐍 Python
• GUI Framework: PyQt5 / Tkinter
• Real-time Visualization: Matplotlib 📊
• AI Models: SOC estimation and temperature prediction using LSTM
• Data Handling: Pandas, NumPy

---

🔄 AI-Based Mode Selection

In this project, we replaced the static threshold-based mode selection logic with a Machine Learning (ML) model. The ML model dynamically selects the optimal mode (Performance, Eco, or Balanced) based on real-time battery conditions, including SOC (State of Charge), temperature, and current.

📊 Data for Mode Selection

The mode selection model is trained on a comprehensive dataset with the following key features:

🔑 Feature	🌟 Description
SOC	State of Charge (%): Indicates the battery's charge level.
Temperature	Predicted Battery Temperature (°C): Key factor for thermal management.
Current	Current Draw (A): Positive for discharging, negative for charging.
FanSpeed	Fan Speed (RPM): Reflects the cooling system's fan intensity.
PumpDutyCycle	Pump Duty Cycle (%): Percentage of time the pump is active for cooling.
CoolingIntensity	Cooling Intensity: Defines the cooling level (Low or High).
Mode	Target Mode: The operational mode (Performance, Eco, Balanced).

🛠️ How These Features Are Used:

• SOC & Temperature: Drive the selection of the most efficient or performance-oriented mode.
• Current: Helps manage discharging/charging scenarios and mode transitions.
• FanSpeed & PumpDutyCycle: Regulate the cooling system to maintain thermal efficiency.
• CoolingIntensity: Adjusts dynamically to match mode-specific requirements.
• Mode: Acts as the target variable for supervised learning.

This dataset enables the mode selection model to make real-time, data-driven decisions for optimizing battery performance, thermal regulation, and longevity.

Here’s a snapshot of the data used to train the mode selection model:

---

📋 Legend

• SOC (%): State of Charge of the battery (in percentage), indicating the battery's available capacity.
• Temperature (°C): Battery temperature in Celsius, critical for thermal management.
• Current (A):
  • Positive values represent discharging (battery providing power).
  • Negative values represent charging (battery receiving power).
• Fan Speed (RPM): Cooling system's fan speed in revolutions per minute, reflecting thermal regulation.
• Pump Duty Cycle (%): The percentage of time the cooling pump is active, significantly influencing heat dissipation.
• Cooling Intensity: Indicates the required cooling level:
  • High: Used in Performance Mode for aggressive cooling.
  • Medium: Balanced cooling for moderate conditions.
  • Low: Energy-efficient cooling, typical for Eco Mode.
• Mode:
  • Performance: Optimized for maximum power output and cooling.
  • Eco: Focused on energy efficiency and extending battery life.
  • Balanced: A trade-off between performance and energy efficiency.

---

🛠️ How the AI Model Works

1. Input Features: The model takes SOC, temperature, and current as input.
2. Prediction: Based on historical training data, the model predicts the optimal mode.
3. Dynamic Adjustments:
   • Cooling intensity and current are adjusted dynamically based on the selected mode.
   • Warnings are generated for exceeding critical thresholds.

---

📈 Advantages of AI-Based Mode Selection

• Dynamic Decision-Making: The model adapts to real-time battery conditions for optimal performance.
• Proactive Adjustments: Predicts potential issues and makes adjustments before they occur.
• Improved Efficiency: Balances battery performance, efficiency, and longevity better than static thresholds.

---

🤖 Machine Learning Integration

This project utilizes Machine Learning (ML) for intelligent mode selection, leveraging a cutting-edge Gradient Boosting Model (GBM) for dynamic decision-making. The model predicts the optimal operational mode—Performance, Eco, or Balanced—based on real-time battery conditions.

📊 Data Utilization

The dataset used for training includes the following features:

• SOC (State of Charge): Battery's charge percentage.
• Temperature (°C): Real-time battery temperature.
• Current (A): Positive for discharging, negative for charging.
• Fan Speed (RPM): Cooling system's operational speed.
• Mode (Target): Labeled modes (Performance, Eco, Balanced) optimized for specific conditions.

🛠️ Mode Selection Model Training

1. Objective:
   • Train a supervised classification model to dynamically select the optimal mode based on real-time battery data.
2. Algorithm:
   • XGBoost (Extreme Gradient Boosting) was chosen for its speed, accuracy, and ability to handle complex, non-linear data relationships.
3. Training Process:
   • A labeled dataset with historical battery conditions was used to train the model.
   • The model's hyperparameters were tuned using grid search to achieve optimal performance.
4. Outcome:
   • The XGBoost model achieved high accuracy in predicting the best mode for various battery conditions.
   • Feature importance analysis revealed that Temperature and SOC are the most influential factors.

🚀 Real-Time Integration

• The trained XGBoost model is integrated into the system to process real-time battery data and predict the best mode on the fly.
• Based on the selected mode:
  • Cooling intensity and current limits are dynamically adjusted.
  • Warnings are triggered when critical thresholds (e.g., temperature limits) are exceeded.

🔍 Benefits of ML-Based Mode Selection

• Dynamic Adjustments: Modes adapt intelligently to match real-time battery conditions.
• Optimized Performance: Balances high power output, energy efficiency, and battery health.
• High Predictive Accuracy: XGBoost provides robust and reliable predictions even with complex datasets.

By leveraging a state-of-the-art Gradient Boosting Model, this project achieves smarter and more adaptive battery management, empowering users with seamless and efficient operation.

---

🧮 Confusion Matrix

The confusion matrix below illustrates the performance of the mode selection model, showing how accurately it predicts each mode based on the dataset:

Purpose:

• Highlights the model's accuracy for each mode (Performance, Eco, Balanced).
• Identifies areas where the model performs well or needs improvement.

📊 Feature Importance

Below is a bar chart showing the importance of each feature used in the mode selection model:

🎯 Decision Boundary Visualization

The scatter plot below illustrates how the model predicts modes based on SOC and Temperature:

📊 Analyzing the Plots

1. Feature Importance Plot

• Temperature: Most critical factor for mode selection, influencing cooling and performance decisions.
• SOC (State of Charge): Second most important, reflecting the battery's charge level and its effect on mode choice.
• Current: Moderately significant, affecting charging and discharging scenarios.
• Fan Speed: Least significant, possibly an indirect predictor.

Insight: Focus on improving data quality for Temperature and SOC, and consider simplifying the model by removing less significant features like Fan Speed.

---

2. Decision Boundary Visualization

• Regions:
  • Pink: Performance Mode.
  • Green: Eco Mode.
  • Blue: Balanced Mode.
• Observations:
  • Clear separation between regions indicates effective classification.
  • Slight overlap near boundaries may lead to edge case misclassifications.

Insight: Optimize the model (e.g., hyperparameter tuning) to improve boundary handling and test with real-world data to ensure robustness.

---

These analyses confirm the importance of the features and demonstrate the model's classification accuracy, reinforcing the ML-driven approach for mode selection.

---

🖼️ Visualization

Below is a snapshot of real-time output based on AI-driven mode selection:

---

📂 Project Structure

AI_BMS_Optimization/
│
├── data/               # Data storage
│   ├── raw/            # Raw input data
│   ├── processed/      # Preprocessed data ready for use
│   └── sample_input.csv # Example data for testing
│
├── src/                # Source code
│   ├── gui.py          # GUI implementation (PyQt/Tkinter)
│   ├── real_time_mode_switching.py # Core script for mode switching
│   ├── models.py       # Includes SOC and temperature prediction models
│   ├── utils.py        # Helper functions (e.g., data preprocessing)
│
├── docs/               # Documentation
│   └── README.md       # Description, usage, and instructions
│
├── tests/              # Test scripts
│   ├── test_models.py  # Unit tests for SOC and temperature models
│
├── requirements.txt    # Dependencies
└── .gitignore          # Ignore unnecessary files

🔄 Modes

Mode	Description
⚡ Performance	Prioritizes maximum battery performance by increasing cooling and fan speed, potentially reducing battery lifespan.
🌱 Eco	Focuses on battery longevity and energy efficiency by lowering cooling intensity and regulating power usage.
⚖️ Balanced	Strikes a balance between performance and efficiency with moderate cooling and power settings.
🛠️ Custom	Empowers users to define parameters like fan speed and cooling thresholds, observing real-time effects.

---

🧠 How It Works

1. Real-Time Data Collection:

   • The system continuously monitors and collects real-time battery parameters, including:
     • 🌡️ Temperature
     • ⚡ Voltage
     • 🔋 SOC (State of Charge)
     • 🔄 Current
     • 🌬️ Fan Speed
     • 💧 Pump Duty Cycle

2. AI-Powered Predictions:

   • AI models predict key battery metrics such as:
     • SOC (State of Charge)
     • Temperature trends
   • Predictions help dynamically optimize battery performance and longevity.

3. Interactive Graphical Interface:

   • The GUI, built with PyQt5/Tkinter, allows users to:
     • Seamlessly switch between modes.
     • Visualize real-time impacts on battery performance via dynamic Matplotlib plots.

---

🤖 Dynamic Mode Switching

• AI-Driven Adjustments:

  • The system automatically adjusts cooling intensity, fan speed, and other parameters based on:
    • The selected mode (Performance, Eco, Balanced, or Custom).
    • Predicted SOC and temperature from AI models.

• Custom Mode:

  • Users can manually fine-tune settings like cooling thresholds and fan speeds.
  • The system provides real-time feedback to help users evaluate their custom configurations.

🚀 Installation

To set up the project locally, follow these steps:

1. Clone the repository:

git clone https://github.com/yasirusama61/AI_BMS_Optimization.git cd AI_BMS_Optimization

2. Install the dependencies

pip install -r requirements.txt

3. Run the GUI: To start the GUI for the mode selection interface, run:

python src/gui.py

Data Used

• Battery Operation Data: Voltage, Current, SOC, Temperature, Pump Duty Cycle, and Fan Speed.
• Environmental Data: Ambient temperature data (Tx) to dynamically adjust system performance.
• Historical Performance Data: Used to train the AI models on performance metrics under different conditions.

📊 Real-Time Output Example

The AI Battery Management System (AI BMS) provides real-time predictions and dynamically adjusts battery parameters based on the selected mode. Below is an example of the output generated during a simulation:

🔍 Highlighted Output

• Mode: Balanced
• Predicted Temperature (°C): Real-time temperature predictions using the AI model.
• Cooling: Cooling intensity automatically adjusted (Low or High).
• Adjusted Current (A): Battery current modified dynamically.
• Warnings: Alerts for exceeding critical thresholds.

---

🖼️ Sample Output

---

🚨 What These Outputs Show

1. Real-Time Adjustments:

   • The system adjusts cooling intensity and current dynamically based on real-time predictions.
   • High Cooling is activated when the predicted temperature exceeds the mode's threshold.

2. Warnings:

   • Alerts (⚠️) indicate when critical thresholds (e.g., temperature limits) are exceeded.

3. Insight into Mode Functionality:

   • The Balanced Mode prioritizes stability with controlled cooling and current.

---

📝 Insights for Developers

• The table demonstrates the AI system's capability to adapt to dynamic battery conditions in real time.
• Warnings help highlight potential issues that require attention, such as exceeding the temperature threshold.

🤝 Contributing

We welcome contributions! Please feel free to fork the repository, submit pull requests, or report issues.